Ransomware at the hypervisor layer: how attackers use ESXi and vCenter features for malicious purposes

Source: Securelist

Author: Elsayed Elrefaei, Mohamed Ghobashy

URL: https://securelist.com/esxi-ransomware/121959/

ONE SENTENCE SUMMARY:

BlackNevas devastated ESXi/vCenter and backups using legitimate interfaces, persistence, Host Profile SSH enablement, credential bridges, and weak logging boundaries.

MAIN POINTS:

  1. GERT observed a 10-day campaign destroying four ESXi hosts, VMs, vCenter, backups.
  2. No hypervisor exploits occurred; attackers used legitimate VMware management interfaces throughout.
  3. Babuk ESXi source leak lowered barriers, enabling many families to build ESXi lockers.
  4. Unsupported ESXi versions increase exposure, even when ransomware targets newer builds too.
  5. ESXi visibility is asymmetric: guest EDR can’t attribute hypervisor-side VMDK modifications.
  6. Common access paths include vCenter vulnerabilities, credential reuse, AD abuse, and backup bridges.
  7. Attackers persisted on VCSA via MeshAgent and Ligolo-ng systemd service masquerading as VMware.
  8. Host Profiles were weaponized to enable TSM-SSH, then vCenter-originated SSH reached ESXi.
  9. ESXi hardening controls worth monitoring: execInstalledOnly, acceptance level, and SSH/Shell state.
  10. Backup infrastructure was intentionally neutralized first, undermining recovery before ESXi detonation.

TAKEAWAYS:

  1. Treat vCenter as Tier 0: isolate identities, networks, and monitor privileged change sequences.
  2. Separate virtualization admin credentials from Windows domain administration to prevent credential pivoting.
  3. Detect correlated behaviors: Host Profile SSH enablement followed by new vCenter-originated SSH sessions.
  4. Make logging independent: remote syslog and preserved scratch storage are essential for forensics.
  5. Engineer recovery as a security boundary with independent access, monitoring, and tested restores.